GitHub Copilot builds Power Automate flows with rough edges
Reviewed by Helen Jones
Power Platform specialist Matt Collins-Jones published a hands-on test on 27 July 2026 showing GitHub Copilot creating and testing Power Automate flows from prompts. The generated flow handled formulas and a connection-dependent condition, but it initially landed outside the intended solution and hit an error in the new designer.
Prompt-led authoring can remove repetitive setup work when a maker needs a basic flow scaffold, formula or connection pattern. That may make small experiments faster for people who already understand the process they want to automate. The test also shows why AI-generated flows still need normal maker discipline. A flow can look correct in the prompt conversation while its solution placement, connections or designer behaviour still need checking before anyone depends on it.
Analysis
Try prompt-led authoring on one non-critical flow, then verify its solution membership, connection references, designer loading and test-run output before enabling it.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "GitHub Copilot builds Power Automate flows with rough edges", Collab365 Spaces. 1 source referenced.